search

chaofengc / PSFRGAN

PyTorch codes for "Progressive Semantic-Aware Style Transformation for Blind Face Restoration", CVPR2021
Other
370 stars 68 forks source link

did you train more big input #41

Closed zhangyunming closed 3 years ago

zhangyunming commented 3 years ago

hi, have you train 1024 input for restore? will 1024 have a better result than 512?

chaofengc commented 3 years ago

I did not have time and resources to train with 1024 inputs. If you want to test, more scale discriminators may help. And you may also need to find proper hyper parameters for the degradation model.